基于长时间序列MODIS影像的黑龙江省主要农作物种植结构变化研究  

Planting Structure Changes of the Main Crops in Heilongjiang Province Based on Long Time Series MODIS Images

作　　者：李晖 陈雨琪 张紫滟 武洪峰[2] 刘占宇[1] LI Hui;CHEN Yuqi;ZHANG Ziyan;WU Hongfeng;LIU Zhanyu(Institute of Remote Sensing and Earth Sciences,Hangzhou Normal University,Hangzhou 311121,China;Information Research Institute,Heilongjiang Academy of Agricultural Reclamation Sciences,Harbin 154005,China)

机构地区：[1]杭州师范大学遥感与地球科学研究院,浙江杭州311121 [2]黑龙江省农垦科学院情报研究所,黑龙江哈尔滨154005

出　　处：《杭州师范大学学报（自然科学版）》2021年第6期98-105,共8页Journal of Hangzhou Normal University(Natural Science Edition)

基　　金：国家自然科学基金项目(41671437);杭州市公益性农发项目(20180432B07).

摘　　要：人口增长及城市化的迅速发展,使得我国的农作物格局发生了重大变化.及时、准确地把握区域主要农作物种植信息,对于制定农业政策、指导农业生产和保障国家粮食安全都具有重要意义.NDVI(normalized difference vegetation index)作为植被提取和地物识别使用最为广泛和有效的植被指数,在农作物信息提取方面具有明显优势.文章以黑龙江省为研究区,选取2001、2010和2017年MODIS(moderate-resolution imaging spectroradiometer)三级产品MOD09Q1,构建250 m分辨率、包含全年46个时相的MODIS-NDVI时间序列数据集,以玉米、水稻、大豆和大小麦4种主要农作物为研究对象,利用时间序列谐波分析法(harmonic analysis of time series,HANTS)对数据集进行平滑处理,通过分析4种作物的时序光谱特征提取生长周期关键阈值,构建不同农作物遥感提取模型,得到2001、2010和2017年黑龙江省主要农作物种植结构空间分布图.结果显示:经过平滑处理重构后的MODIS-NDVI时间序列集能较好表现农作物的物候特征;构建的基于MODIS影像的农作物分布提取模型能较好地对农作物进行提取,与黑龙江省历年农业统计年鉴数据对比,平均精度达到87.4%.研究表明所采用的方法适用于大区域、大尺度农作物提取,能为农业遥感监测和农作物种植结构调整等提供有效可行的参考和借鉴.The population growth and the rapid development of urbanization cause the major changes of the crop pattern in our country.Timely and accurate grasp of regional main crop planting information is of great significance for formulating agricultural policies,guiding agricultural production and ensuring national food security.NDVI,as the most widely used and effective vegetation index for vegetation extraction and feature recognition,has obvious advantages in crop information extraction.This article takes Heilongjiang Province as the research area and selects the MODIS three-level product MOD09Q1 in 2001,2010 and 2017 to construct a MODIS-NDVI time series data set with a resolution of 250 meters and 46 phases throughout the year,taking corn,rice,soybean and big wheat as the research objects,using harmonic analysis of time series to smooth the data set,extracts the key thresholds of the crop growth cycle through analyzing the time series spectral characteristics of these four crops in Heilongjiang Province,and establishes the remote sensing extraction models of different crops to obtain the main crops planting structure spatial distribution map in Heilongjiang Province in 2001,2010 and 2017.The results show that the MODIS-NDVI time series set after smoothing and reconstruction can better represent the phenological characteristics of crops.The construction of a crop distribution extraction model based on MODIS image can better extract crops,comparing with the Heilongjiang Provincial Agricultural Statistics Yearbook data,the average accuracy reaches 87.4%.The method used in this paper is applicable to large-area and large-scale crop extraction,and can provide an effective and feasible reference for agricultural remote sensing monitoring and crop planting structure adjustment.

关 键 词：MODIS NDVI 长时间序列 种植结构 

分 类 号：S127[农业科学—农业基础科学] TP79[自动化与计算机技术—检测技术与自动化装置]